An Iron(III) Superoxide Corrole from Iron(II) and Dioxygen

Abstract

A new structurally characterized ferrous corrole [Fe^II(ttppc)]⁻ (1) binds one equivalent of dioxygen to form [Fe^III(O₂^•−)(ttppc)]⁻ (2). This complex exhibits a ^16/18O₂-isotope sensitive ν(O–O) stretch at 1128 cm⁻¹ concomitantly with a single ν(Fe–O₂) at 555 cm⁻¹, indicating it is an η¹-superoxo (“end-on”) iron(III) complex. Complex 2 is the first well characterized Fe-O₂ corrole, and mediates the following biologically relevant oxidation reactions: dioxygenation of an indole derivative, and H-atom abstraction from an activated O–H bond.

Graphical Abstract

An iron-superoxide corrole has been generated from a ferrous corrole and dioxygen, and was found to be capable of indole dioxygenation and H-atom transfer reactions.

The activation of dioxygen by metalloenzymes usually involves one-electron transfer from a reduced metal center (Mⁿ⁺) to O₂ to give a metal-superoxo (Mⁿ⁺¹(O₂^−•)) species. A subset of these species involves iron enzymes, and there is significant interest in the biological roles of iron-superoxo intermediates, including their proposed involvement in substrate oxidations.^[1] This reactivity has been proposed for heme enzymes such as tryptophan/indoleamine 2,3-dioxygenases (TDO/IDO) (indole dioxygenation),^[2] MarE (indole monooxygenation),^[3] and nitric oxide synthase (NOS),^[4] and nonheme iron enzymes such as isopenicillin-N-synthase (hydrogen atom transfer (HAT)),^[5] myo-inositol oxygenase (HAT),^[6] and cysteine dioxygenase (S-oxygenation).^{[1b, 7]} Metal-superoxo species are also proposed as key intermediates in a number of synthetic, oxidative catalytic cycles.^[8]

The inherent instability of Fe^III(O₂^−•) species makes their preparation and direct characterization a challenge. Iron-superoxo porphyrins derived from O₂ have been known for some time,^[1a] and recently, analogous nonheme Fe^III(O₂^−•) species have been reported.^[9] For the porphyrin complexes, studies describing the oxidative reactivity of the FeO₂ adducts are beginning to emerge,^[10] but these data remain relatively scarce, and the key factors that influence this reactivity (e.g. bond strengths, spin state, redox potentials, pK_a) remain under significant debate.

Corroles are a remarkably versatile family of tetrapyrrolic compounds that have been employed in a wide range of applications, including catalysis, as synthetic models of metalloenzymes,^[11] and as artificial cofactors for heme proteins (e.g. myoglobin, peroxidases).^[12] Many of these applications involve oxidative transformations, and the reactivity between metallocorroles and oxidants such as O₂ or related analogs (e.g. H₂O₂, ROOH) comprises a critical step in their mechanisms of action. The activation of O₂ with biologically relevant iron complexes is anticipated to involve an initial FeO₂ adduct. However, essentially nothing is known about FeO₂ corroles. We set out to conclusively identify such a species with a sterically bulky corrole ligand, and provide the first measure of its oxidative reactivity.

Herein we report the synthesis and characterization of an iron(III)-superoxo corrole, prepared from a new Fe^II corrole precursor and O₂. The superoxo complex was characterized by UV-vis, Mössbauer, ¹H NMR, and resonance Raman spectroscopies, and to our knowledge, is the first example of a well characterized Fe(O₂)(corrole) complex.^[13] This complex is capable of indole dioxygenation and H-atom abstraction, two fundamental processes proposed to be mediated by biological Fe(O₂) species.

Reduction of brown Fe^III(ttppc)^[11c] (ttppc = 5, 10, 15-tris((2,4,6-triphenyl)phenyl)corrolato(3-)), with sodium/mercury amalgam (1.5% Na in Hg) gives [Fe^II(ttppc)]⁻ (1). Single crystal X-ray diffraction reveals a 4-coordinate ferrous corrole, [Na(H₂O)(2-MeTHF)(1)] (Figure 1). A sodium cation is weakly bound to an N_pyrrole atom (Na–N_pyrrole = 2.553(5) Å) and the nearest phenyl ring (Na–C28 = 2.861(7) Å). There is one H₂O and one 2-MeTHF molecule bound to the Na⁺ ion. The Fe–N bond distances range between 1.862(4) – 1.893(4) Å, and the iron atom lies along the plane of the macrocycle (Fe-to-mean plane(23 atom core) = 0.044 Å). These metrical parameters are close to those reported for [Fe^II(tpc)]⁻, the only other structurally characterized Fe^II corrole.^[13] Complex [Na(H₂O)(2-MeTHF)(1)] provides a rare example of a square-planar Fe^II compound.^[14] Magnetic susceptibility measurements for 1 (Evans Method) gave μ_eff = 3.44 ± 0.2 μ_B, higher than the theoretical value for an S = 1 ion (μ_eff(spin-only) = 2.83 μ_B), but similar to other Fe^II (S = 1) complexes.^[15]

Figure 1.

Displacement ellipsoid plot (50% probability level) for [Na(H₂O)(2-MeTHF)(1)] at 110(2) K. Hydrogen atoms omitted for clarity.^[27]

The UV-vis spectrum of 1 (Figure 2) reveals distinct peaks at λ_max (ε in M⁻¹ cm⁻¹) = 465 (56300), 545 (15200), 588 (15700), and 645 (29200) nm, and the ¹H-NMR spectrum of 1 in THF-d₈ exhibits relatively sharp, paramagnetically-shifted peaks between +16 to −83 ppm (Figure S2). The zero-field Mössbauer spectrum of an ⁵⁷Fe-enriched sample of 1 reveals a single, sharp quadrupole doublet with δ = 0.25 mm s⁻¹ and |ΔE_Q| = 2.45 mm s⁻¹ (Figure 2). There are no reported Mössbauer spectra of other iron(II) corroles for comparison, but the isomer shift of 1 is comparable to other ferrous (S=1) porphyrin/porphyrinoids.^[16]

Figure 2.

(a) Generation of 2 from 1 + O₂ in THF at −80 °C. (b) Titration of 1 (24 μM) with O₂ (0 – 1.0 equiv) at −80 °C. (c) Overlay of zero-field ⁵⁷Fe Mössbauer spectra for 1 and 2 in THF at 80 K. Global fits for 1 (green) and 2 (brown).

Exposure of 1 to air in THF at 23 °C results in immediate oxidation back to Fe^III. However, reaction of 1 with excess O₂ at −80 °C results in isosbestic absorption changes and new peaks at λ_max = 454 nm (70800 M⁻¹ cm⁻¹), 595 nm (20500 M⁻¹ cm⁻¹) assigned to 2. Spectral titrations support a 1:1 Fe:O₂ stoichiometry (Figure 2b). The new species exhibits less than 10% decay over 1 hour at −60 °C, but complete decay to Fe^III(ttppc) within two hours at −40 °C. These observations suggest an assignment for 2 as a superoxo complex, [Fe^III(O₂^•−)(ttppc)]⁻.

Zero-field Mössbauer spectroscopy of 2 at 80 K reveals a single, sharp quadrupole doublet with δ = 0.10 mm s⁻¹, |ΔE_Q| = 3.45 mm s⁻¹ (88% of the total fit) (Figure 2c). A similar quadrupole doublet was observed upon lowering the temperature to 4 K (Figure S4). The lack of hyperfine splitting at 4 K is consistent with an integer-spin species, as seen for other iron-superoxide porphyrins.^{[17] 1}H NMR (−80 °C) spectra of 2 gives only poorly defined peaks in the diamagnetic region (Figure S5).

The assignment of 2 as an Fe^III(O₂^−•) species was solidified by resonance Raman (RR) spectroscopy. RR spectra of 2 at 110 K were obtained with a 458-nm laser excitation and showed intensely resonance enhanced bands attributable to corrole macrocycle vibrations (Figure 3). Comparing RR spectra of samples prepared with ¹⁶O₂ and ¹⁸O₂ identifies a strong band at 555 cm⁻¹ that down shifts to 531 cm⁻¹ with ¹⁸O-labeling. These frequencies and the 24 cm⁻¹ downshift are consistent with a ν(Fe–O) stretching mode assignment. Another band observed at 1128 cm⁻¹ in the spectrum of 2 prepared with ¹⁶O₂ and absent in its ¹⁸O₂-counterpart falls within the expected range for a ν(O–O) stretching mode of mononuclear η¹-superoxo (end-on) metal complexes.^[18] The isotope-edited difference spectrum supports this assignment and reveals that the downshifted ν(¹⁸O–¹⁸O) mode expected around 1064 cm⁻¹ (diatomic harmonic oscillator approximation) affects the intensities of strong corrole vibrations at 1043 and 1069 cm⁻¹. Similar mode mixing and intensity borrowing processes have been observed with porphyrin iron-superoxo complexes.^[18c–e] These RR spectra identify 2 as an iron(III)-superoxide species. The high ν(Fe–O₂) and low ν(O–O) frequencies sustain the back-bonding correlations previously observed for XO adducts of Fe(II) porphyrins.^[19] Moreover, while a side-on superoxo species would be expected to display two Fe–O₂ stretches (i.e. symmetric and asymmetric stretches), the concomitant detection of a single ν(Fe–O) at 555 cm⁻¹ with a ν(O–O) at 1128 cm⁻¹, and with intensities and frequencies that are similar to those observed in the RR spectra of end-on iron(III)-superoxo tetrakis(2,6-difluorophenyl)porphyrinate and yeast oxyhemoglobins,^[20] support an end-on binding geometry of the superoxo ligand in 2 along with the presence of a weakly coordinating THF as a sixth ligand.

Figure 3.

Resonance Raman spectra of 2 prepared with ¹⁶O₂ (black) and ¹⁸O₂ (red) in 2-MeTHF at 110 K (λ_exc = 458 nm); difference spectrum (¹⁶O₂–¹⁸O₂) (green) and was multiplied by a factor 4 above 900 cm⁻¹ to better visualize the 1128-cm⁻¹ band and accompanying perturbations near 1064 cm⁻¹ where the ν(¹⁸O-¹⁸O) is expected to occur.

We set out to establish the oxidative capabilities of 2 in two biologically relevant reactions: indole dioxygenation and H-atom abstraction. The initial step of indoleamine oxidation in TDO/IDO is postulated to involve an Fe^III(O₂^•−)(porphyrin) intermediate, although direct evidence for this step remains scant.^{[10c, 21]} Addition of 2,3-dimethylindole (2,3-DMI) to 2 in THF at −40 °C gives an immediate color change from brown-green to red, and analysis by TLC and ¹H NMR spectroscopy identified N-(2-acetyl-phenyl)-acetamide as the major product (86%)(Figure 4). When O₂ was bubbled into a solution of 1 preloaded with 2,3-DMI at −40 °C, similar results were obtained. The use of ¹⁸O₂ leads to a prominent ion at 181 m/z in the electron impact mass spectrum, which corresponds to the incorporation of two ¹⁸O atoms in the dioxygenated product (Figure S7). A peak at 179 m/z is also observed, corresponding to the incorporation of a single ¹⁸O, suggesting that there is one or more intermediates that can undergo exchange with an ¹⁶O source (e.g. trace H₂O).

Figure 4.

a) Reaction of 2 with 2,3-dimethylindole. b) Time-resolved UV-vis spectra (0 – 45 min) for the reaction between 2 (60 μM) and 2,3-DMI (100 equiv) in THF at −40 °C. Inset: changes in A_{595 nm} over time. c) Plot of pseudo-first-order rate constants (k_obs) vs [2,3-DMI] together with best-fit line.

It is known that free superoxide anion reacts with tryptophan derivatives to give ring-opened, dioxygenated products.^[22] Given that 2 decays slowly to an Fe^III product at −40 °C, we considered the possibility that there may be some free superoxide in solution, or there may be a rapid dynamic equilibrium between bound and free O₂^•− to give a steady state amount of O₂^−•, which in turn could react directly with 2,3-DMI. A control reaction was run involving addition of a KO₂ mixture to 2,3-DMI under the same conditions as employed for oxidations with 2. No appreciable formation of any oxidation products was observed (Figure S8). This control reaction rules out participation of free O₂^−• in the oxidation of the indole substrate.

Monitoring the reaction between 2 and 2,3-DMI by UV-vis showed the characteristic peak for 2 (595 nm) converts to a peak at 575 nm assigned to Fe^III(ttppc) following first-order kinetics, with isosbestic points at 580 and 664 nm (Figure 4b). Varying the substrate concentration led to a second-order rate constant k₂ = 0.02(0) M⁻¹ s⁻¹ (Figure 4c). The lack of an observed intermediate suggests that attack of 2 on 2,3-DMI is the rate-determining step. Subsequent steps may follow those proposed for the enzymes, although the enzymatic mechanisms are still not well understood and remain a topic of significant debate.^[23] Proposed mechanisms for certain TDO enzymes also suggest that the initial reaction between the putative Fe^III(O₂^•−) species and the substrate is the rate-determining step.^[2c]

Initial efforts to determine the ability of 2 to abstract hydrogen atoms began with relatively weak C-H substrates, e.g. cyclohexadiene (CHD). However, no reaction was observed. Reaction of 2 with the O–H bond donor TEMPOH led to an isosbestic conversion to Fe^III(ttppc) (Figure 5), and TEMPO radical (97%), consistent with a hydrogen atom transfer (HAT) reaction. Attempts to detect the H₂O₂ product via an iodometric method^[24] were unsuccessful.

Figure 5.

a) Reaction of 2 with TEMPOH. b) UV-vis spectra (0 – 30 min) for 2 (19 μM) and TEMPOH (75 equiv) in THF at −40 °C. Inset: changes in A_{454 nm} over time. c) Plot of pseudo-first-order rate constants (k_obs) vs [TEMPOH] (green circles) or [TEMPOD] (purple circles) together with best-fit lines. d) Eyring plot.

Kinetic analysis shows pseudo-first-order behavior with excess TEMPOH, yielding k_2,H = 1.03 M⁻¹ s⁻¹ at −40 °C. Deuterated TEMPOD gave k_2,D = 0.205 M⁻¹ s⁻¹ and a KIE(k_H/k_D) = 4.21(Figure 5c). An Eyring analysis yielded activation parameters of ΔH^‡ = 11.0 kcal mol⁻¹ and ΔS^‡ = −11.9 kcal mol⁻¹ K⁻¹ (Figure 5d). The data clearly point to a bimolecular reaction involving hydrogen atom transfer between TEMPOH and 2.

The extrapolated k₂ for 2 at −80 °C is 0.016(0) M⁻¹ s⁻¹, which is ~30 to 130-fold slower than those reported for Fe^III(O₂^•−)(porph) complexes.^{[10a, 10d]} The steric encumbrance of the meso substituents in 2 could be a factor in slowing H-atom abstraction, but previous examination of M(OH)(ttppc) (M = Fe, Mn) showed that these complexes reacted with O–H substrates at rates similar to other less hindered M(OH)/(O) complexes.^[11e]

A control reaction with free KO₂ was carried out for TEMPOH, and did show production of TEMPO radical (Figure S13). This result was not unexpected given the relative ease of oxidation of TEMPOH. At this time we have no evidence for the presence of free superoxide in these reactions, but we cannot rule out the participation of free O₂^•− in the oxidation of TEMPOH. We note that the earlier reports on metal-superoxide reactivity with TEMPOH do not appear to have addressed this issue,^{[10a, 10b, 10d, 25]} but the possible involvement of free O₂^•− in the mechanism of HAT does need to be considered for metal-superoxide species.

In summary, we report the first well-characterized iron(III)-superoxo corrole via dioxygen activation by a ferrous corrole precursor. The ferrous corrole is only the second example to be structurally characterized, and confirms a rare, square-planar iron(II) configuration for this complex. The spectroscopic characterization of 2 provides benchmarks for identifying other Fe^III(O₂^•−) corroles, such as in artificial proteins or oxidation catalysts. Complex 2 was capable of indole dioxygenation, adding to the very few examples of iron-superoxide complexes^{[10c, 21]} shown to carry out the initial step hypothesized for certain TDO/IDO enzymes. These results support the proposed enzyme mechanism, but also indicate that dioxygenation of indole by Fe^III(O₂^•−) is possible in the absence of an optimized enzyme/cofactor active site.^{[10c, 21]} Complex 2 was also competent to oxidize TEMPOH by HAT, although significantly slower than a porphyrin analog. The factors that influence the dramatic differences in HAT reactivity for metal-superoxide species remain poorly understood,^{[25a, 26]} and future studies should help enlighten researchers in this arena.